🐣 Download Hay Day for free and get 16 exclusive decorations to personalise your farm—worth $15.99! It’s available until the 15th of February! #HayDay https://link.haydaygame.com/?action=voucher&code=04493209-0183-48f3-87d7-2e582db465b1
In recent years, the internet has been inundated with a huge amount of AI generated “history” videos with yassified portrayals of historical people and questionable-at-best information. What exactly are we up against, and how do these generative AI systems even work? Come learn with me!
Thank you to Ryan Rowe
Karolina Żebrowska’s video: https://www.youtube.com/watch?v=yMsCbdDKnEI&pp=ygUVa2Fyb2xpbmEgemVicm93c2thIGFp
Care for Gaza: https://x.com/careforgaza?lang=en
Donate to the PCRF: https://www.pcrf.net/
Buy my graphic novel, Liberated: The Radical Art and Life of Claude Cahun: https://www.kazrowe.com/comics/cahun
https://shop.getty.edu/products/liberated-the-radical-art-and-life-of-claude-cahun-978-1947440074
Find me on Twitter: https://twitter.com/KazRowe
I’m on TikTok @ kazrowe
Find me on IG: https://www.instagram.com/kaz.rowe
Buy my comics: https://gumroad.com/kazrowe
Send me a ko-fi: https://ko-fi.com/A347K4X
Read my webcomic: https://www.cunningfire.com/
Catch it on Tapas: https://tapas.io/series/cunning-fire
Line Webtoons: https://www.webtoons.com/en/challenge/cunning-fire/list?title_no=59016
Captioning by Transcription, Ho! Captioning Services
Filmed using:
Blackmagic Pocket Cinema Camera 4k– https://www.blackmagicdesign.com/products/blackmagicpocketcinemacamera
Blackmagic Video Assist 5” HDR – https://www.blackmagicdesign.com/products/blackmagicvideoassist
Olympus M. Zuiko ED 12-40mm f/2.8 Pro– https://www.bhphotovideo.com/c/product/1003635-REG/olympus_v314060bu000_m_zuiko_digital_ed.html
Samsung Portable SSD T5 – 2Tb– https://www.samsung.com/us/computing/memory-storage/portable-solid-state-drives/portable-ssd-t5-2tb-mu-pa2t0b-am/
~~~~~~~~~~~~~~~~~~~~~~~~
Sources
“AI models struggle with expert-level global history knowledge” by Eric W. Dolan For PsyPost https://www.psypost.org/ai-models-struggle-with-expert-level-global-history-knowledge/
“AI is creating fake historical photos, and that’s a problem” by Marina Amaral for The Colour of Time Blog https://marinaamaral.substack.com/p/ai-is-creating-fake-historical-photos
United States District Court Southern District of New York: Roberto Mata v. Avianca Inc. https://storage.courtlistener.com/recap/gov.uscourts.nysd.575368/gov.uscourts.nysd.575368.31.0.pdf
“Turns Out, I’m Not Real: Detecting AI-Generated Videos” by Bernadette Young for Columbia Engineering https://www.engineering.columbia.edu/about/news/turns-out-im-not-real-detecting-ai-generated-videos
“The potentially dangerous implications of an AI tool creating extremely realistic video” by William Brangham and Karina Cuevas for WTTW PBS News Hour https://www.pbs.org/newshour/show/the-potentially-dangerous-implications-of-an-ai-tool-creating-extremely-realistic-video
“People Are Creating Records of Fake Historical Events Using AI” by Chloe Xiang for Vice https://www.vice.com/en/article/people-are-creating-records-of-fake-historical-events-using-ai/
“Is AI Ruining Our Perception of History?” by Karolina Żebrowska on Youtube https://www.youtube.com/watch?v=yMsCbdDKnEI
“‘Historical Figures’ AI Lets Famous Dead People Lie To You” by Miles Klee for Rolling Stone https://www.rollingstone.com/culture/culture-news/historical-figures-ai-chat-bot-lies-dead-people-1234664257/
source
🐓 Get 16 free decorations to personalise your farm – available until the 15th of February! Download Hay Day now by clicking here. #HayDay
https://link.haydaygame.com/?action=voucher&code=04493209-0183-48f3-87d7-2e582db465b1
This is not a dig at your cousin, I have classmates and friends (in animation) who also have a more positive outlook on AI but I have a hard time genuinely considering AI as a potential tool for "progress" when all the ways it can be abused clearly outweighs any benefits. It's not a matter of developing it better, the dangers of AI are inherit to it. Although I appreciate you considering his perspective because not everyone who's lulled into AI are doing it to be greedy or lazy, you kinda just have to say out right that if AI was ethical, it'd be entirely redundant to the big tech companies pushing it…
The message of this video was so beautifully and passionately presented, you did a really wonderful job! I already agreed with you on all points before watching this, but I truly hope you swayed some minds and made someone reflect who really needed it.
Also, really enjoyed the talk with Ryan, that was great.
Also also, your tattoo is beautiful 😊
Ryan's bias is very palpable every time he talks
As a historian it's fucking infuriating. Recently I came across a FB post about the Trail of Tears with clearly AI-generated pics of a group of Natives trudging in the middle of a creek in the snow… with perfectly open spaces all around. Like what the fuck? And of course everyone fell for it as if it were the real thing. It's incredibly depressing.
A lot of the comments are negative towards your cousin, so I’ll not add to that. I think it’s interesting that he says he wants to work with AI to make it better when the rise of AI is why I left computer science. I came to the grim realization that there are very few fields that aren’t tainted by the desire for capital, but technology is one of the worst. I just wanted to help make websites for people to get away from relying on social media…. (That’s still something I can do without a degree and my current degree program suits me much better.)
I noticed it's already starting to get the hands right, somehow in the last few weeks a lot of the inconsistancies went away
As an artist, generative AI needs to be nuked from orbit
I honestly think this video was harmed by the interview section; he really just threw out corporate buzzwords and didn't seem to have a grasp on how machine learning algorithms actually WORK or what they can do, let alone the boons and harms they do to society, particularly online. He was just repeating talking points someone else drilled into his head.
Your cousin is very confidently incorrect.
Giving my experience as an art student (recently graduated). While I never saw anyone use ai for art homework, my creative writing teacher mentioned seeing it a lot, one key point being how ai would not include dialogue unless prompted
I recently interned at an art museum where an artist used generative AI as part of their work. The artwork was all about the future of our oceans so it felt both on point and a bit hypocritical to be using (given the environmental impact). I’m unsure what the future of AI is in the museum space and how I feel about it (specifically as artworks)
I've been a long time fan of the channel and this video was very strange to me. I understand the constraints and rough n' tumble of video production but as someone with a background in computer science and who worked with PhD students with a focus on machine learning the Ryan section was difficult to watch. I wish him the best, but as others have commented he has certainly seemed to have been taken by corporate A.I propaganda, or is trying to protect his potential employment (understandable).
Edit: Man I gotta admit the further discussion about being excited for the use of "A.I" bummed me out. We have had access to these dumb algorithms for a very long time at this point and they are no closer to becoming intelligent. The nomenclature is purely for advertising purposes. I apologize if this seems mean spirited, I just want more from the people who are already so knowledgeable in so many things. Machine learning shouldn't be such a sticking point. The only use of machine learning I have found and thought to be both ethical and helpful was a project utilizing image recognition tech to categorize galaxies.
Edit Edit: Oh I am losing my mind. Kaz oh god no the problem with LLMs is not that they do not understand nuance, they understand nuance as well as anything else, which is to say they understand nothing at all!!! Kaz I am begging you an LLM will only do as well on a test with nuanced questions as it has datasets answering similar questions oh god we cannot evaluate an A.I on how well it understands nuance this is a horrible metric to base your claim on!!! Kaz nooooooo!!! Think of the reverse hypothesis!!! If it could understand nuance it would still not be able to explain or understand history ethically or well, entirely intrinsically to it's design!!! (This section of the video is still strong however, Kaz does go over how machines cannot process new information properly, will only ever dispense what info they've been trained on, biases from who created it (not exactly but right ballpark, machine learning will often pick up new biases undiscovered from datasets like the tuberculosis incident) etc.)
Edit Edit Edit (Final edition): Idk I think we do need to hold trusted educators to a higher standard???? A historian should know this????
Edit Edit Edit Edit (Final edition 2): Fuuuuck and referencing Karolina's amazing video on the subject?? what are we doing???
Thank you. This was well done.
That interview with your brother was a great explanation of things.
AI generation of architectural images, particularly historical images is pure crap. It consistently gets it—not just wrong—but comically, horrendously wrong.
I'm so angry at all the old media my relatives got rid of. All that pre-ai information gone. I want to scream.
What happens when the US users ask the Chinese folks on Little Red Book about Tianamen Square, though? Or what life is like for 2nd children born under the one child policy? I was hoping you might cover those aspects of revisionist history, censorship etc. when talking about the migration. I think the "normalcy" or similar aspects of Chinese society shared on the app belie the much more subtle ways the CCP controls access to media and information and how their polices impact life in China. Misperceptions of Chinese society have continued to exist not just because US education is shitty (though I have my doubts most people were listening in history class and I find it hard to believe every school district was just doing straight Red Scare BS well into the 2000s as peoples responses would imply) but is also because their limited access to many of the websites most of the rest of the world uses to share "slice of life" types of content. They were not even on the same iteration of TikTok as the rest of the world.
I love history, but I hate colorized historical film. For one thing, there’s hardly any color. Every automobile is a constantly shifting pattern of purple and gray.
AI images have made Pinterest basically unusable
Been a quiet subscriber of yours for a handful of years now but wanted to speak up to say: Thank you, Kaz. Thank you so much for continuing to make videos, for citing sources, for taking the time we spend with you so seriously and putting so much effort into your videos. I've never felt like a moment spent watching your videos has been a single moment wasted. Those who don't give your videos the proper time don't realize what they're missing out on. I've always appreciated the energy, flow, tone, and cadence to your videos–they've always held my interest and attention and strike the perfect balance of informative and respectful of our time, in my opinion.
I've always admired your approach to the foggier parts of history, where you've mentioned, "we can't know for sure, but. . ." or "we can't really coin or qualify historical figures by the standards and terminology of today but. . ." as well! I get settled in and interested in learning as soon as I start up one of your videos. Hearing your, "Come learn with me," feels more and more like a friend inviting me to take a seat and discover something new alongside them, rather than be lectured at and warms my heart and gets me excited to see what you'll bring up.
I started to tear up as you talked about, "Are we okay with rocketing towards a future where we don't ever talk to or hear each other?" and tackled "If we just use A.I. to do everything for us, then what's the point?" and appreciate that you allowed yourself a moment to recognize you were getting emotional, too. It made me feel less alone in how much I find myself tangled up in caring, too.
Thank you, Kaz, for giving a fuck and for taking the time to do what you do and bring us your genuine work and educate us from a place that recognizes nuance, that comes from a genuine intrigue in the subject you're discussing, and takes the time to be respectful of the sources and those you're communicating this knowledge with. Please keep on doing what you're doing. We need you and folks like you now more than ever. ❤️
AI has done ABSOLUTELY NOTHING to improve our lives. it's ruining them. I can't wait for all the money invested in it getting sucked into a sink hole of wasted time. there's better things to spend money on than this BullDust. they're only putting money into it because the aren't bothered to help with humanitarian issues
As a person studying about the world, AI slop has always crossed my feed to the point it's hard to stay on certain apps for a long period of time. I only use YouTube now, but AI content has grown exponentially. It's getting hard to watch news about geopolitics and getting bombarded with slop.
Kaz, I'm a huge fan of your work and have been for ages now, but I can't finish this video. The interview section felt like I was watching a billionaire talking to investors, and that just turned me off. Then you started talking about how "AI" will be used to detect cancer. No, you're completely and absolutely wrong and it sounded like you failed to understand what this technology actually is. That, coupled with being preached to about how these massive companies are actually being very careful is horrifying. I don't think you'll take this video down because it's genuinely unwatchable, but you should.
I watched a history person watch a course on YouTube and it objectively broke down why it's not right.
I need to say something about the tattoo, it looks nice but it needs to be a whole sleeve now I feel, I'd like to see your teddy bear and a clock with the time sticks on a spring yo represent this channel and the was you show different times and you dress the part.
I also forgot to tell you the episode where you was in Scotland was wonderful your naming of places was perfect anc better than people from off the bbc.
Sorry for the essay, it's way psst my bed time.😊
I study art history and fine arts y so I've yet to see anything AI generated in critiques, but last year there was a kid who would pay other people to paint their paintings for them. they would come to critique like "no conceptual meaning. I just think its pretty." made me want to scream
00:30. As a WHAT tool!?
I have listened to some AI stories on YouTube recently. A lot of it is absurdly bad. They are repetitive, both of each other and within each single story. They will change the pronouns of the person and other basic grammar and consistency errors. And then you get absurd similes like "her skin was silky, like a piece of silk."
Artificial intelligence is not very intelligent so far.
My social studies teacher showed us a video about ancient Egypt that me and my classmates could tell was ai generated, the teacher also made us talk to a king tut ai chatbot to ask him about ancient Egypt. It honestly made me pretty uncomfortable because we were being showed ai stuff instead of learning material made by humans 🙁
Hey kaz, idk if you’ll see this or not but I love your videos! I always watch to the end, even if I’m just listening to them while doing other activities; I DO learn from you and trust that you have done your research and had fun while doing so.
Anyway thanks for making content ❤
Your videos and others like it (that is to say, well researched, informed and nuanced historical videos) are my preferred dopamine hit 🎯
Can I get ONE FUCKDAMN VIDEO THAT DOESN'T MENTION THAT TRAITOR?!!!!
AI will likely be the end of the internet as most people know i. It will devolve in to AI talking to AI reacting to AI and no normal person will bother using it beyond what is needed for work.
If we don't pull back from it, we will lose so much.
AI should help humanity with mass data analysis, medical pattern analysis, etc.
Not spread misinformation or steal from artists.
how are the capchas training ai if it already knows the correct answer? i don't think that's right
Going to strongly disagree that ai colorization can replace human colorization. History requires a lot of knowledge and context that a historian would have but an ai is physically incapable of replicating. We should Never settle for less accurate sources when we have other options, the only possible reason I could think to justify this is cutting financial corners, and when you have as much money as these institutions thats inexcusable. Outside of very specific scientific cases the only real big change AI has made is shrinking the job market significantly. Megacorps have shown time and time again that quality means nothing, only the numbers. I firmly believe the world would've been better off if this was an obscure tool only accessible to researchers and never released to the public. We already have our more accurate, quality sources at our finger tips
"open AI is pretty transparent" oh, honey…
26:20 You know that video is fake because Simon Cowell does not smile.
Great video! Many thanks for this. You just got a new subscriber (not for the following reason but rather your intellectual rigour).
Also you are permitted to tell your cousin that a random internet stranger declares him "hot as fuck". If he doesn't like swearing you are permitted to censor it. Thank you.
As a teacher, when I start writing an assignment into MS Teams it automatically offers to improve it with AI. So, i am supposed to make sensless AI assignments the students will answer with AI and the AI will grade it?
i watch your videos through and throughout 🧛🏼🧛🏼 i feel so educated after watching you and your so cool dont stop living 🙏🏼
Interview ends at 16:19
i really like how people shared what was happening in Palestine like how they keep attacking Israel despite being oppressed little know fact victims of holocaust actually instigated the Nazis into taking more extreme actions in 1940….pretty much is what people are saying
If no regulation, then the best case we get is rehashes till we d!e
Basically the only 'new' films would be "Star Wars 10,11,12,13, spin off 10, spin off 12" "Cinderella 2,3,4,5" and so on
Like you say, it needs data to refer to before anythings spit out.
Meaning new content from unknown artists can't be trained off, since the chance of them being able to pay rent is basically nill.
If the only ones making money are big companies like Disney, milking every cow dry, to the point where the Sahara starts to sweat.. No new things will be made ever again.
As the opportunity to make money (and even pay bills) through making new things is taken away.
If I used chatgpt to create my grocery list for later today, this would probaby become the funniest day ever